Anthropic opens Singapore office with AI safety MOU

Anthropic opened a Singapore office on 14 October and signed a memorandum of understanding with Singapore’s Ministry of Digital Development and Information. The company says the agreement will cover AI safety, security and model evaluation, research on AI adoption, and responsible use across public and private sectors.[1]

The announcement extends a company presence and policy partnership in a nearby Southeast Asian market. It does not establish a Malaysian office, commit the parties to a specific procurement, or publish a funding amount or delivery timetable. The practical value will depend on the exchanges and programmes that follow.[1]

The MOU was signed on the sidelines of Singapore International Cyber Week by Permanent Secretary Kai Fong Chng and Anthropic co-founder Ben Mann, with Singapore’s digital development minister and Anthropic’s APAC policy head as witnesses, according to the company. The announced areas include AI safety and security, model evaluation, adoption-impact research and responsible adoption. A Ministry of Home Affairs transcript dated 13 October separately records that Minister K Shanmugam met Anthropic and other major firms at the conference. It corroborates official engagement around SICW, but does not document this MOU.[1][2]

Anthropic says it plans technical exchanges with Singapore institutions on model evaluation, testing standards and incident response. It also says the parties will look at AI skills, support for small and medium-sized businesses, public-officer capability, research sharing and policy expertise as Singapore develops AI rules. These are stated areas of work; the announcement does not give milestones, budgets or public deliverables.[1]

Anthropic says the Singapore office is its fifth in Asia-Pacific and that it already serves government and business customers there. Its release reports that more than 1,500 engineers and technical staff at GovTech Singapore have adopted Claude Code. The Cyber Security Agency of Singapore is exploring possible use of Claude for cyber defence; exploration is not deployment.[1]

In finance, Anthropic says DBS uses Claude for software development and productivity within its security and governance framework, while GIC is rolling the tool out across investment and corporate-services teams. The company also says Grab’s BriX sales assistant saves about 40,000 staff hours each quarter. These examples are company-reported and do not include independent measurements or full cost comparisons.[1]

The office gives Anthropic a local base in a market where the company says Claude use is high. Its latest Economic Index places Singapore second among 121 countries for Claude.ai usage relative to population. That ranking is Anthropic’s measure of activity on its own service; it is not a measure of total AI adoption or business productivity across Singapore.[1]

For Malaysian organisations, the announcement is a regional signal rather than a new local service commitment. The MOU names support for SMEs and public agencies in Singapore, but does not extend those programmes to Malaysia. Firms comparing AI suppliers should still ask where data is processed, what controls apply, how systems are evaluated and what evidence supports claimed gains.[1]

The announcement also does not say whether cooperation will involve access to model weights, what information the parties may exchange, or how responsibilities will be divided when a technical issue is found. Without the MOU text, readers cannot assess the precise obligations or dispute process. These are material details for work involving safety evaluation and incident response.[1]

For Malaysian firms that use regional cloud or AI services, the Singapore office does not by itself change a Malaysian contract, data location or support commitment. Buyers should confirm those terms with each provider and assess whether a regional service route fits their own privacy and security requirements.[1]

The release does not publish the MOU text, an implementation schedule, evaluation methods or a list of specific joint projects. It also gives no eligibility rules for SME support or training. Until those details appear, the agreement is best understood as a framework for cooperation rather than a finished programme businesses can apply to.[1]

The next evidence to watch is a named technical exchange, a published evaluation or incident-response protocol, and concrete SME or public-service projects with results. Anthropic says the office is open and hiring, but does not state how many staff it will add. Those future details will show how the policy agreement translates into local capability.[1]